NVLink Fusion joins Arm Neoverse for next-gen AI data centers
Arm and NVIDIA are extending their long-running collaboration with the integration of NVIDIA NVLink Fusion into the Arm Neoverse platform. The announcement that underscores the accelerating shift toward energy-efficient AI data center architectures. As hyperscalers race to boost AI throughput without escalating power budgets, the companies aim to deliver a unified, coherent infrastructure capable of scaling to next-generation workloads.
For eeNews Europe readers working as system architects, chip designers, and data-center engineers, this development offers insight into how future AI platforms will be built. It appears new architectural options are on the table for coherent CPU-accelerator integration and rack-scale system design.
Expanding Neoverse for the AI era
Arm noted it is positioning Neoverse as the compute foundation for power-efficient, highly scalable AI deployments. With more than 1 billion Neoverse cores already deployed and broad hyperscaler adoption from AWS, Google, Microsoft, Oracle, and Meta, the platform is on track to reach 50% market share across top cloud providers in 2025. Major AI data-center initiatives (such as the Stargate project) are also committing to Arm-based compute for energy-efficient scaling.
Against this backdrop of rising demand, Arm says it is extending Neoverse with NVIDIA NVLink Fusion to bring Grace Blackwell-class performance, bandwidth, and coherency to the broader ecosystem. NVLink Fusion provides a high-bandwidth, coherent pathway between CPUs, GPUs, and accelerators, enabling partners to build custom, rack-scale AI systems with flexible accelerator choices.
“Arm and NVIDIA are working together to set a new standard for AI infrastructure,” Rene Haas, CEO, Arm, noted. “Extending the Arm Neoverse platform with NVIDIA NVLink Fusion brings Grace Blackwell-class performance to every partner building on Arm — a milestone that reflects the incredible momentum we’re seeing in the data center.”
Jensen Huang, founder and CEO of NVIDIA, added: “NVLink Fusion is the connective fabric of the AI era — linking every CPU, GPU and accelerator into one unified rack-scale architecture. Together with Arm, we’re extending this vision across Neoverse to empower innovators everywhere to design the next generation of specialized AI infrastructure.”
Coherent integration for high-bandwidth AI systems
The companies highlight that NVLink Fusion is designed to interface directly with AMBA CHI C2C, Arm’s coherent, high-bandwidth chip-to-chip protocol. By enabling Neoverse with the latest C2C specification, Arm indicated that it ensures seamless integration between Arm-based CPUs and accelerators connected over NVLink Fusion.
This compatibility gives ecosystem partners a coherent fabric for moving data efficiently across heterogeneous AI systems, reducing memory bottlenecks, accelerating system development, and cutting time-to-market for emerging AI accelerators.
The approach builds on previous achievements such as the Grace Hopper and Grace Blackwell platforms, where tight CPU-GPU coherency set new benchmarks for high-performance computing. With NVLink Fusion now available across the full Neoverse ecosystem, Arm and NVIDIA are widening access to these capabilities, enabling differentiated system designs optimized for intelligence per watt.
A strengthening partnership
As AI models grow in size and power budgets tighten, the Arm–NVIDIA partnership seems to be evolving toward deeper co-design at the platform level. By combining Arm’s energy-efficient compute with NVIDIA’s high-bandwidth interconnects, the companies look to be aiming to define the next generation of AI data-center architecture.
For AI hardware developers and data-center operators in Europe, this integration signals a new phase of scalable, energy-efficient infrastructure — one where coherent CPU-accelerator designs become the norm rather than the exception.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
